Image Classification on MNIST (Gauss, SUAP, GSUAP, VIP)
92.6GaussGSUAP
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GSUAPModel=ResNet-50, Ratio (%)=202026.05 | 92.6 | 16.7 | 16.3 | 99.8 | |
| GSUAPModel=ResNet-18, Ratio (%)=202026.05 | 49 | 10.2 | 10.2 | 99.6 | |
| GSUAPModel=ViT-Tiny, Ratio (%)=202026.05 | 46.3 | 11.2 | 10.4 | 99.3 | |
| GSUAPModel=ResNet-50, Ratio (%)=402026.05 | 45.7 | 11.1 | 10.9 | 99.8 | |
| GSUAPModel=ViT-Small, Ratio (%)=202026.05 | 40.8 | 10.8 | 10.7 | 99.2 | |
| GSUAPModel=ResNet-50, Ratio (%)=602026.05 | 35.3 | 10.4 | 10.2 | 99.8 | |
| GSUAPModel=ResNet-50, Ratio (%)=802026.05 | 33.7 | 10.4 | 10.2 | 99.8 | |
| GSUAPModel=ViT-Tiny, Ratio (%)=402026.05 | 32.4 | 10.9 | 10 | 99.3 | |
| GSUAPModel=ResNet-50, Ratio (%)=1002026.05 | 32 | 10.4 | 10 | 99.8 | |
| GSUAPModel=ResNet-18, Ratio (%)=402026.05 | 31.3 | 11.3 | 11.3 | 99.6 | |
| GSUAPModel=ResNet-18, Ratio (%)=602026.05 | 29.4 | 9.3 | 9.3 | 99.6 | |
| GSUAPModel=ViT-Small, Ratio (%)=402026.05 | 26.1 | 10 | 10 | 99.2 | |
| GSUAPModel=ResNet-18, Ratio (%)=802026.05 | 24.7 | 10.6 | 10.2 | 99.6 | |
| GSUAPModel=ViT-Tiny, Ratio (%)=602026.05 | 23.8 | 9.4 | 8.9 | 99.3 | |
| GSUAPModel=ViT-Small, Ratio (%)=602026.05 | 22.7 | 9.8 | 9.4 | 99.2 | |
| GSUAPModel=ResNet-18, Ratio (%)=1002026.05 | 22.6 | 10.2 | 10.2 | 99.6 | |
| GSUAPModel=ViT-Tiny, Ratio (%)=802026.05 | 21.2 | 9.9 | 8.4 | 99.3 | |
| GSUAPModel=ViT-Tiny, Ratio (%)=1002026.05 | 19 | 9.8 | 8.4 | 99.3 | |
| GSUAPModel=ViT-Small, Ratio (%)=802026.05 | 17.9 | 9.5 | 9.5 | 99.2 | |
| GSUAPModel=ViT-Small, Ratio (%)=1002026.05 | 14.1 | 9.5 | 9.4 | 99.2 | |
| GSUAPModel=AlexNet, Ratio (%)=202026.05 | 12.6 | 9.8 | 9.7 | 98.9 | |
| GSUAPModel=DenseNet, Ratio (%)=202026.05 | 12.3 | 10.7 | 9.5 | 99.5 | |
| GSUAPModel=AlexNet, Ratio (%)=402026.05 | 11.5 | 9.4 | 9.5 | 98.9 | |
| GSUAPModel=AlexNet, Ratio (%)=602026.05 | 10.9 | 9.2 | 9 | 98.9 | |
| GSUAPModel=AlexNet, Ratio (%)=802026.05 | 10.6 | 9.2 | 8.9 | 98.9 | |
| GSUAPModel=AlexNet, Ratio (%)=1002026.05 | 10.6 | 9.2 | 8.9 | 98.9 | |
| GSUAPModel=DenseNet, Ratio (%)=802026.05 | 10.5 | 9.5 | 9.5 | 99.5 | |
| GSUAPModel=DenseNet, Ratio (%)=1002026.05 | 10.4 | 9.4 | 9.4 | 99.5 | |
| GSUAPModel=DenseNet, Ratio (%)=602026.05 | 10.3 | 10.3 | 9.3 | 99.5 | |
| GSUAPModel=DenseNet, Ratio (%)=402026.05 | 10.1 | 9.8 | 9.8 | 99.5 |